person view; (e) editing in 3D; (f) walking the landscape
2Beginning. Regardless of whether the contract is the result of a competition,
advertisement, or a direct negotiation, the start of every planning and design process is marked by detailed exchange of information between the landscape planner and the client. It is of particular importance at this stage that comprehensive landscape data is passed on to the landscape planner. The planner often does additional mappings. The more precise the data, the more potentially accurate the three-dimensional landscape visualizations will be.
3Initial state—GIS. A standard tool used by landscape planners is the Geographic Information System (GIS), which allows the profile of the landscape and the
environment to be summarized and represented. Using GIS, the spatial data can be used to create digital maps. GIS provides landscape architects with powerful tools for storing, analysing and modelling the spatial data. The GIS themes necessary for three-dimensional visualization with the Lenné3D-System are a topographical map, a digital elevation model (DEM) and other thematic maps, such as a habitat map.
4Initial state—Lenné3D. The GIS data are loaded into the LennéD–MapEditor. The LennéD–MapEditor forms an interface between the spatial data and the interactive three-dimensional visualizations; the landscape planner uses it to direct the individual visualization projects and representation settings. This software component unites the advantages of conventional modelling with the classical two-dimensional map, in a digital form. The digital model of the terrain is processed into an interactive, dimensional visualization on the basis of the GIS data. Additionally,
three-dimensional objects such as buildings can be loaded and landmarks and labels can be added. The data are summarized in a Lenné3D project. The interactive three-
7.9 Screenshot from the Lenné3D-MopEditor; blending of an aerial photo and habitat map (ESRI Shapefile), both draped on a digital terrain model
dimensional maps provide an intelligible alternative to traditional maps, which demand a high level of abstraction. Yet they also use the clearness of traditional illustrations for editing or adding other geographical or landscape data. Special illustrative techniques such as information lenses allow a selective and clear communication of information (Figure 7.9).
Applications in the agricultural landscape 139
5Virtual walk. The clearness of the three-dimensional map representation in the LennéD–
MapEditor allows those involved in planning to decide on a starting point or route for a walk through the landscape model, which they can explore as a photorealistic, virtual-reality model using the Lenné3D–Player.
6Plant distribution. With the help of a further component called oik—Lenné3D’s vegetation modeller and plant distributor – GIS data are used to calculate rule-based distribution of plants for the most natural possible simulation of a real landscape. To this end, habitats are broken down into various types of vegetation. The plants are then distributed randomly within the vegetation layers and units. Mapped plants (e.g. a solitary tree) are positioned interactively or according to precise coordinates from a GIS point theme. Without this software component, the vegetation in the simulated landscape would be unnaturally regularly positioned. The landscape architect controls the oik component through the 3D–MapEditor.
7Plant modelling. The plant models used in the visualization afterwards are three-dimensional textured models, created using the software Xfrog (Lintermann and Deussen 1998). They correspond to biologically correct criteria and, after conversion, allow a smooth level of detail and real-time representation. The user will have the option of selecting from a large number of ready-made plant models.
8 First person vieiv. The Lenné3D–Player allows the observer to navigate through landscape section selected in the 3D–MapEditor, interactively, in real time and with a breathtaking degree of realism, all from the perspective of someone walking through the landscape.
9 Planning with GIS. Following the inventory and analysis, discussions with the client and other stakeholders, and the preliminary walk through the digital landscape model, the landscape planner proceeds to develop a planning concept for the landscape. The objectives for nature protection and landscape management and of alternatives are saved as GIS data. Requirements and proposed measures for the client and stakeholders are derived from the planning concept.
10 Editing in 3D. The GIS data are processed for interactive use in the three-dimensional visualization by the Lenné3D–MapEditor. The distribution of plants is calculated for the types of habitats envisaged. The landscape architect uses the 3D–MapEditor for analysis and the planning of modifications that will affect the scenery. The editing function with three-dimensional visual control allows different versions to be designed, all of which may be saved as GIS data.
11 Walking the landscape. Whilst walking through the interactive landscape with those involved in the landscaping process, specific suggestions may be taken on board;
should there be aspects of the design that are unclear, it is possible to turn back for a second look, in order to gain a precise picture of the situation. Various different design options, planning alternatives or scenarios of landscape development can be shown successively or in parallel for the sake of comparison. Virtual measures can be
selected or modified within the available decision space. If further modifications to the planning are necessary, the iterative process is repeated from stage 9.
Methods
Using real examples of landscape planning, the research and development team is engaged in testing application of the prototype system and developing the process of stakeholder participation supported by computer graphics. The first trial run started in early 2003 within the framework of the project ‘Interactive Landscape Plan’ in Königslutter (www.koenigslutter.de/landschaftsplan.htm), sponsored by the German Federal Office for Nature Conservation (BfN). The aim of the collaborative project is to test Lenné3D in practice under the conditions of a participative landscape planning process, and to evaluate its acceptance and results in an accompanying survey. In order to guarantee development of a system that will stand up to professional use, landscape planners are directly involved in the application design of the software. Furthermore, specialist seminars with potential professional users are conducted regularly.
Contrary to the linear software development originally envisaged, an iterative, incremental procedural model is the chosen method of approach. This model corresponds to contemporary standards in software engineering and is particularly suitable as a development tool for distributed projects with complex software systems.
The software architecture of the system designates two main components: the 3D–
MapEditor for the assembly, exploration and editing of landscape data, and the 3D–
Player for the interactive landscape visualization from first person view. This partitioning arises from the implementation of specialized and optimized computer-graphics systems, developed as separate projects, for each of the two components. Both of these main components employ the oik component, developed for the generation of plant distribution.
The 3D-MapEditor is based on the authoring software for three-dimensional maps and three-dimensional city models LandEx (www.landex.de), which is being developed further for application in the field of landscape visualization. Analysis and editing functions, which serve to ascertain ecological parameters, have been integrated into the 3D–MapEditor especially to fulfil the requirements of vegetation modelling. With the 3D–MapEditor, Lenné3D projects are created and managed; the component can also integrate and transform geo-data.
The 3D–Player is a program that interprets the scenic description and geo-data specified in a Lenné3D project. Three-dimensional vegetation models are assigned to the plant distribution calculated by the oik component and positioned within the landscape model. At the forefront of this technology is the development of the Level of Detail (LoD) process for the dynamic reduction of the overall complexity of the scenery. The geometric and colour complexity of the landscape model must be drastically reduced in order for the real-time simulation in the 3D-Player to be possible, but this must remain imperceptible to the viewer. This requires that the relevant visual information is conveyed and packaged in suitable data structures for efficient data processing and quick access. The latest computer graphics techniques allow a large number of images to be generated and calculated per second. Triangles of the complex model are progressively replaced with fewer primitives (points and lines) as the camera moves away.
Representing object surfaces as sets of points or lines without connectivity allows for easier simplification and generation of LoD representations. The point and line algorithm
Applications in the agricultural landscape 141
projects each point or line sample onto the screen and draws it as a pixel. At scene level, whole objects are progressively ‘melted’ into the terrain texture as the camera moves back.
The development and implementation of illumination procedures for plant and vegetation models involve some special features with regard to prior illumination methods. For example, in certain situations, the light is dappled as it shines through the plants or their leaves. The illumination model must fulfil this additional requirement.
Special new graphic hardware and programming capabilities are utilized in the 3D–
Player for a flexible implementation of such illumination models, so that they can be adapted to various user requirements.
The oik component calculates plant positions by means of heuristicalgorithmic models, on the basis of habitat data (biotope types), reference mappings and topographical site data. In cooperation with the company Greenworks Organic Software, best known for its Xfrog software, a wide range of central European plant life, particularly foliage, has been modelled. Plants commonplace in central Europe are selected, and supplemented by species typical of the region in which the testing areas are situated. The three-dimensional plants are modelled for the 3D–Player in full spring appearance. This decision was made for reasons of working capacity, as well as the fact May-June seems the most suitable phenological reference season for visual simulation, as the vegetation is most lush at this time. Woody plants are also modelled for a winter view.
First results
Initial stages of the project aimed to construct a basic framework for the development and integration of the software components of the Lenné3D system. The intention was to integrate the different, hitherto rudimentary, software components at an early stage of the project, in order to avoid a ‘big-bang’ integration in the final stages—which, in the field of software engineering, carries a greater risk than an iterative, incremental approach.
A prototype system is already being tested with GIS data sets. A preliminary version of the 3D–Player can already be used to visualize smaller landscape scenes photo-realistically and in real time (Figure 7.10).
In collaboration with our partner project ‘Interactive Landscape Design’, a 1.1 km2 slope was selected for high-detailed landscape visualization, in discussion with local stakeholders. The area, which is predominantly used for agriculture, is situated at the edge of the Elm Natural Park, Germany and is affected by erosion. The town wishes to enhance soil protection, promote landscape-related recreation and protect the natural species and habitats it houses.
Conclusion
The profession of landscape planning ‘must be prepared to keep up to date with current developments in the field of digital technology and, if necessary, develop solutions tailored to its needs’ (Rekittke 2002:121). After two years of development, the project’s goal of real-time visualization of complex landscape scenery is deemed realizable considering the performance that is already possible using current three-dimensional
graphics cards, as well as the increasing performance made possible by developments in hardware and corresponding algorithmic innovations. Nevertheless, an effective solution has to be found for processing the considerable amounts of data involved in large and complex landscape scenes with millions of plants.
7.10 Screenshots from the interactive Lenné3D–Player, geodata from the Uckermark district: (top left) existing;
(top right) as planned with removal of farm building; (bottom left) and (bottom right) higher plant density, more species and more than 2.5 million three-dimensional plants instances (bottom images reproduced in the colour plate section)
But how will landscape planners and stakeholders react to the new technological possibilities, especially in controversial situations? This new form of communication could stimulate interest in civic participation, while promoting a fun, educational approach to environmental protection—thereby generating more interest among young people and members of the population who have so far shown only scant interest. Three-dimensional landscape visualization is by no means a typical service of landscape designers. Until now it belonged to the domain of the film and computer-game industries.
Yet the influence of new media technology is reshaping our visual habits, and along with Applications in the agricultural landscape 143
these the demands on and potential for presentational illustrations of our environment.
Among landscape planners and their clients there is increased awareness of the benefits of landscape simulation during the planning and design process, not purely as a medium for presentation, but as an integral and integrating component of a participative design process, in a departure from the classical, linear model.
Architects and town planners have always constructed three-dimensional models (either physical or virtual); now, landscape planners will also be able to use visual simulation to add a three-dimensional aspect to planning. Visualizations from first person view using GIS data are coming into conflict with the scale of conventional representations of community landscape plans. A greater degree of detail can call into question the accuracy of the basic data and the postulated planning scale of, for example, 1:10 000.
As more ‘bottle-necks’ are removed by the fast pace of developments in computer graphics, the more transparent typical issues of landscape design such as cost, availability, quality and incongruities of data and models will become apparent.
The research project Lenné3D has set itself high aims. In part they are so high that they cannot be completely fulfilled during the term of the project, yet it is already manifestly apparent that the research is leading to developments that will considerably improve future methods of communication and visualization during the planning and design process. Our strategy of holding bi-annual seminars with professional experts and users to hear their assessments and criticisms of latest developments in the research project has proven extremely valuable, as well as demonstrating that our results have already inspired confidence and enthusiasm among user groups.
It becomes apparent ‘that landscape, by virtue of its inherent intellectual or virtual nature, is ideally suited to being experienced and conveyed through digital media’
(Rekittke 2002:110). However, an important revelation for all those involved in the project was the realization that the reality of natural, botanical and scenic beauty could not be even closely approximated through computer simulation. Rather than a source of disappointment, we see this as a continuing challenge for our current and future work.
In addition, the current rapid technological developments and our own development milestones have served to increase our respect for the skill of conventional methods of representation. Our chosen name Lenné3D serves as a constant benchmark for checking and recalibrating the quality of our graphics.
Acknowledgements
The German Federal Environmental Foundation (DBU) sponsors the Lenné3D project.
Lenné3D is a cooperative project of the Institute of Land Use Systems and Landscape Ecology (ZALF), the Zuse-Institut Berlin (ZIB), Department of Scientific Visualization, the University of Konstanz, Deptartment of Computer and Information Science, Chair Media Informatics, Prof. Dr Oliver Deussen, and the Hasso-PlattnerInstitute for Software Systems Engineering (HPI), Chair Computergraphical Systems, Prof. Dr Jürgen Döllner.